Summary

We started this chapter by looking at design considerations for a graph database pipeline, and we also refamiliarized ourselves with how to set up a Neo4j graph database. Our use case for this chapter was creating a graph database for retail, and we designed a schema and pipeline. With our schema considerations mapped out and considered, we then looked at how you can add static data and introduced fake data to simulate customer interactions. Obviously, this would not be fake in practice but served as a good way to test out if our desired schema functioned the way we would want it to in a production environment.

The ultimate aim of this chapter was to set up a schema that would enable us to make product recommendations based on similar products customers buy. The first step was to get refamiliarized with Cypher (Neo4j’s query language – similar to SQL) and Python for working with this data, followed by making recommendations by brand. This then led on to recommendations...